Carefree maneuvering using adaptive neural networks

This paper describes carefree maneuvering techniques for aircraft under multiple limit constraints on multiple control axes. Adaptive multi-layered neural networks are employed for on-line learning and modelling error compensation. The approach utilizes an observer type adaptive neural network loop for an estimation of the correct aircraft model. The identified aircraft model is then used to predict the dynamic trim response behavior of the limit parameters. The corresponding control margins are calculated for each control corresponding to each different limiting value. A penalty vector for each control channel can than be calculated based on the control margins. Only standard sensor measurements are used for adaptation. The effectiveness of the proposed adaptive limit detection technique is evaluated through a series of simulations using the nonlinear Generic Tiltrotor Simulation (GTRSIM) program.